Capacity of Group-invariant Linear Readouts from Equivariant Representations: How Many Objects can be Linearly Classified Under All Possible Views?
Matthew FarrellBlake BordelonShubhendu TrivediCengiz PehlevanPublished in: ICLR (2022)
Keyphrases
- invariant representations
- object models
- similar objects
- d objects
- database
- camera views
- multiple views
- probability models
- multiple objects
- viewing angle
- object description
- video sequences
- geometric transformations
- moving objects
- object representations
- topological features
- planar objects
- invariant properties
- invariant representation
- object views